Misfire Analysis in Combustion Engines Using Vibration Signals Based on the Calculation and Reduction of Statistical Indicators

Abstract

In this research work, a diagnosis methodology based on the calculation and reduction of statistical features that are obtained from vibration signals is proposed for the detection of misfire or spark failure in an internal combustion engine (ICE). This work is carried out by the characterization of vibration signals using four statistical features that are capable of modeling trends and describing changes in the acquired vibration signals. In addition, the proposed method introduces the consideration of dimensionality reduction techniques such as principal component analysis and linear discriminant analysis, which allows to obtain the reduction of an original feature space leading to achieve a visual representation of the characteristic patterns, represented by the set of statistical features, that characterizes different evaluated conditions in the ICE. The proposed method is evaluated on experimental data acquired during normal and misfire operation of an ICE in low and high speed regime. The processing of the acquired vibration signals and the application of the proposed diagnosis methodology is performed under Matlab®, which is a sophisticated software that may be used in a wide range of engineering applications. The results obtained in two-dimensional characteristic patterns clearly show the characterization of different operating conditions evaluated in the ICE, these results demonstrate that the methodology allows detecting faults in an ICE that are generated in the ignition system and that the detection of faults can be carried out effectively regardless of the rotational speed of the engine.